Priya Kohli

Priya Kohli

Assistant Professor of Statistics

Joined Connecticut College: 2012

M.S., Indian Agricultural Statistical Research Institute, New Delhi
M.S., Northern Illinois University
Ph.D., Texas A&M University


Temporal, spatial and spatio-temporal modeling

Covariance modeling

Longitudinal studies

Multivariate data analysis

Missing data

RNA-seq analysis

Financial applications

Priya Kohli specializes in the areas of covariance modeling, longitudinal/panel studies, multivariate modeling, missing data, time series, spatial statistics, and spatio-temporal modeling. She also works in interdisciplinary research areas including RNA-seq analysis, healthcare devices, environmental sciences, and business and finance.

Priya Kohli Curriculum Vitae‌ (pdf)

Priya Kohli Resume (pdf)

Professor Kohli’s research accomplishments include publication in the time series book titled Economic Time Series: Modeling and Seasonality, her research work accepted as a U.S. patent and also recently filed as a European patent, and her publications in some of the most distinguished international journals. She has been invited as a speaker to present her research work at several prestigious international conferences, including the Joint International Chinese statistical Association and Graybill Conference, National Bureau of Economic Research – National Science Foundation Time Series Conference, Second Conference of the International Society of Nonparametric Statistics in Cadiz, Spain, International Symposium on Business and Industrial Statistics, International Conference on Advances in Interdisciplinary Statistics and Combinatorics. Previously she taught Statistical Methods and Principles of Statistics at both Texas A&M University and Northern Illinois University.

More recently, she has been working on developing methodology for statistical discrimination of nonlinear and nonstationary time series. This is an important problem in many diverse areas such as financial markets, biomedical studies, and environmental sciences, when some or all of the time series exhibit nonlinearity and/or nonstationarity. It turns out that when some subset of the series exhibits nonlinear characteristics, frequency domain clustering methods are based on higher order spectral properties, such as the bispectra or trispectra. For this reason there are several open problems related to higher-order properties, including the estimation of time-varying, higher order spectra for non-linear time series, and the development of optimal clustering algorithms based on these spectral estimates. She is working on addressing some of these challenges.

Kohli is also currently working on developing a unified framework for modeling dependence structure in multivariate longitudinal studies for complete and incomplete data. Her research focuses on the developing data-adaptive unconstrained parametrizations for parsimonious modeling of covariance matrices to address the two major challenges in modeling covariance, that is, high-dimensionality and positive definiteness constraint.

Kohli teaches Statistical Methods, Advanced Regression Techniques, Time Series Analysis, Probability, and Mathematical Statistics.

Kohli has been serving as a statistical consultant for the Committee on Faculty Compensations at the College. In this role, she has worked closely with the other members of the committee on issues related to assessment and effectiveness of the existing salary model(s), equity, and gap-closing.

Recent publications:

Online work:

Kohli recently collaborated with John C Cangelosi '15 and Jill Whitney (Class of 1984) on analyzing the data collected by a survey of young adults on matters related to how parents address the issue of talking with kids about sexuality and values. The major findings based on this collaboration are reported at

Recent talks:

  • Covariance Modeling of Multivariate Longitudinal Data with Application in Clinical Trials, Celebrating Statistical Innovation and Impact in a World of Big & Small Data, IISA, December 20 - 24, 2015.
  • Clustering Time Series: A PSLEX-Based Approach at 24th ICSA/Graybill Joint Conference, Fort Collins, Colorado, June 14-17, 2015.
  • Technology: An Indispensable Tool for Teaching Statistics in the 21st Century, 27th International Conference on Technology in Collegiate Mathematics, Las Vegas, Nevada, March 12-15, 2015.
  • Time Series Clustering, International Conference on Advances in Interdisciplinary Statistics and Combinatorics, October 10-12, 2014.
  • Prediction of Stationary Random Fields with Quarter-Plane Past: A Time-Series Approach, International Society of Nonparametric Statistics Conference, Spain, June 12-16, 2014.
  • Prediction of Stationary Random Fields with Quarter-Plane Past: A Time-Series Approach, International Society of Nonparametric Statistics Conference, Spain, June 12-16, 2014.
  • Restricted Linear Covariance Models for Multivariate Longitudinal Data, NRC Statistics and Probability, Harvard University, July 2014.
  • Clustering Financial Time Series, International Symposium on Business and Industrial Statistics, Duke University, June 9-11, 2014.

Workshops Conducted

  • Introductory R Workshop for Applications in Life Sciences, Quantitative Life Sciences Program, Connecticut College, July 6-7, 2015.
  • Recent advances in statistical techniques for researchers in Biology, Botany and Neuro-science Quantitative Life Sciences Program, Connecticut College, January 5-7, 2014.

Visit the department of mathematics website. 

Majoring in Mathematics. 

Contact Priya Kohli

Mailing Address

Priya Kohli
Connecticut College
Box #5422
270 Mohegan Ave.
New London, CT 06320


312 Fanning Hall